Jump to Body Text

Intervention for Men Who Batter: A Review of Research

Richard M. Tolman
University of Michigan

Jeffrey L. Edleson
University of Minnesota

Published: June 2, 1995


This article is in S.R. Stith & M.A. Straus (1995) (Eds.) Understanding partner violence: Prevalence, causes, consequences and solutions(pp. 262-273). Minneapolis, MN: National Council on Family Relations.


In the 1970 s a resurgent women s movement focused attention on the lack of resources available to battered women, and the lack of sanctions against men who raped or beat their female partners as the most dramatic example of how women s rights were violated in our society ( Schecter, 1982). Intervention with male batterers developed as part of this larger movement addressing the rights and needs of battered women.

In 1977, eight men who were friends of women s activists in the Boston area came together to form a men s collective called EMERGE ( Adams & McCormick, 1982). They began to provide services to batterers. Their creation of services for batterers responded to the frustration of shelter workers who noted that women were being beaten when they returned home, and to the observation that some men were moving from one violent relationship to another. EMERGE became one of the first organizations to offer group treatment for men who batter. The growth of services for batterers rapidly followed the founding of EMERGE ( Feazell, Myers, & Deschner, 1984; Pirog-Good & Stets-Kealey, 1985).

During the same period that batterers programs were being established, the dramatic proliferation of family treatment models resulted in attempts at counseling couples and families with men who abused their partners ( Coleman, 1980; Cook & Frantz-Cook, 1984; Geller, 1982; Geller & Walsh, 1977-1978; Margolin, 1979). The use of these models for intervention with batterers has created a great deal of controversy, however, family and couple treatment interventions constitute a small minority of interventions discussed or evaluated in the research literature.

As a result of lawsuits against police, pressure from women s and victim s advocates, and research evidence, many criminal justice systems began to change their response to battering. Police began to arrest batterers in increasing numbers. Currently, a large number of batterers appearing in court are mandated to batterer treatment. In this context, batterers treatment is one component of a coordinated community response to battering, which includes the criminal justice system response as well as services for battered women and their children. Community intervention projects (CIP) seek to coordinate the response of these systems to provide sanctions for men s violent behavior and to coordinate needed services for victims and their families ( Pence, 1983, 1989; Soler, 1987; Edleson & Tolman, 1992).

These three types of intervention with men who batter-men's groups, couples counseling, and community intervention projects-dominate in North America. Despite the growing use of both criminal justice and social service interventions, the evidence for their efficacy is not clearly established. This paper reviews the research on effectiveness of social service interventions for batterers. However, the effectiveness of these services must be viewed in the context of the other responses which may impact upon batterers abusive behavior. Because many men who batter come to intervention as a result of arrest and prosecution, it is important that services not be viewed in isolation from this context. Therefore, these areas are briefly reviewed below.

Effectiveness of Criminal Justice System and Coordinated Interventions


Until recently, arrest has been widely accepted as the most effective police action in domestic disturbance cases. This conclusion was partially based on the results of Sherman and Berk's ( 1984) highly publicized experiment that found arrest almost twice as effective as other actions in reducing recidivism among violent men. More recent studies have failed to replicate this finding. Funded by the National Institute of Justice, these studies sought to correct some of the weaknesses of Sherman and Berk's design and to identify other factors that may contribute to successful police interventions in domestic violence cases. In three communities, Omaha, Nebraska, Charlotte, North Carolina, and Milwaukee, Wisconsin arrest was not found to be more effective than other options, including advising or separation ( Dunford, Huizinga and Elliot 1990; Hirschel, Hutchison and Dean 1992; Sherman, et al. 1991). In two other communities, Colorado Springs and Metro-Dade, Florida arrest was shown to have deterrent effects ( Sherman 1992). In some cities, among some offenders, arrest was associated with increased long-term recidivism ( Sherman, 1992). However, where increases were found they appeared only in official records, not in reports by subsamples of victims ( Sherman, 1992). In the Omaha replication, Dunford ( 1992) showed that tracking of offenders over longer periods of time yielded increased rates of recidivism. In short, as Sherman ( 1992) states, after more than a decade of evaluating arrest for misdemeanor domestic violence, we still have much to learn" (p. 42).These findings have accelerated the debate on whether arrest is an effective deterrent for domestic violence.

The finding that arrest alone does not consistently lead to lower official recidivism must be interpreted cautiously. For example, Sherman et al. ( 1991) point out that only 5 percent of the men in their study were charged by the courts and only 1 percent were convicted. Studies examining arrest alone have not taken into account the combined effects of coordinated community efforts, including prosecution, harsher penalties, and victim advocacy.


The prosecution rates of domestic violence cases have been typically low ( Dutton 1988; Ford 1983), which may have made police more reluctant to arrest perpetrators because they believed that the work invested in making an arrest would not lead to further legal action ( Dutton, 1987; Ferraro & Pope 1993). Victims viewed police and prosecutors as unwilling to protect them, so they feared increasing the batterer s rage by pressing charges ( Ferraro & Boychuk, 1992; Jaffe, Hastings, Reitzel & Austin, 1993). Consequently, all the parties involved seemed to be reinforcing one anothers inaction.

The frequent classification of battering as a misdemeanor, even when the conduct is very serious, may suggest to victims that the behavior is being trivialized" and that it is not worth the risks of prosecution for such minor consequences ( Hart 1993). Long delays, lack of witness protection, and prosecutors seeming indifference can discourage victim cooperation. The process of prosecution may be so difficult for victims, that some experts maintain that a no-drop policy may deter victims from reporting crimes. By reporting abuse, victims may sense that they have lost control of the legal process ( Buzawa & Buzawa 1990). Cahn argues that jurisdictions with no-drop policies must have victim protection programs ( Cahn 1992).

One recent study examined the connection between prosecution and recidivism. Ford and Regoli ( 1992) used official records and victim interviews to compare outcomes on four prosecutorial tracks involving 678 cases. In cases where women initiated the complaint, they were permitted to drop charges, but charges could not be dropped if they resulted from an officer-on-the-scene arrest. Recidivism rates were determined by court records, personal interviews with victims and defendants, and court observations. A six- month follow-up was included. Under the no-drop" prosecution track, 20% of the cases were still dropped, because the victim refused to testify or disappeared. In cases of victim initiated complaints, giving the victim a chance to drop the charges was most effective in reducing recidivism. According to victim reports, prosecutorial action of any type lowered the risk of recurring violence 50% within six months of case settlement when compared to the level of preprosecution violence ( Ford & Regoli 1992).

Ford ( 1995) argues that interventions must be focused on improving responses to the victim. He points out that victims are likely to use criminal justice proceedings in ways that may not have been intended yet are equally successful from the victim s point of view. For example, he argues that victims may use the threat of prosecution to negotiate their security but not intend to follow through with prosecution. This highlights the importance of considering how the goals of victims for system response may be very different than the goals of other system participants.

Coordinated Efforts

The above studies only allow indirect inferences concerning the impact of coordinated community responses. Steinman ( 1988) evaluated a coordinated community intervention in Nebraska and found that post-arrest sanctions had little influence on recidivism beyond what was achieved through the initial arrest. However, his data suggested that certain post-arrest sanctions (e.g., fines) might be associated with lower rates of recidivism. To investigate these possible associations, Steinman ( 1990) conducted a second study that compared cases before and after implementation of the coordinated community effort. He found that police actions that were not coordinated with other sanctions led to increased violence. Police action, especially arrest, in coordination with other criminal justice efforts became a significant deterrent. Similar findings were reported in a recent study of 425 women who had been contacted by the Minneapolis police concerning a woman abuse report. Syers and Edleson ( 1992) found that police visits to the home, combined with an eventual arrest of the perpetrator, which was also followed by court-mandated treatment, were significantly more likely than other combinations of criminal justice actions to end repeat incidents of violence.

Jaffe, Hastings, Reitzel, and Austin ( 1993) conducted a study in Ontario, Canada, of 90 women who were victims of battering in 1988-89 after a new mandatory arrest protocol was established. Nearly 60% (n=52) had police intervention and pressed charges, fourteen had intervention and no charges, and 24 had no police intervention or charge. Jaffe et al. found that police intervention and charges decreased violence and that women were generally satisfied with the police response. Women also increased their use of social services after the target incident, because they became more aware of these services after police intervention.

While we have much to learn about the effectiveness of criminal justice response, we also have much to learn regarding other types of social interventions with abusive men. We now turn our attention to the issue of the effectiveness of those interventions.

Batterer Intervention Groups

Theoretical Orientation

The majority of published articles describing batterers treatment utilize some variation of cognitive-behavioral or social learning approach. Many programs incorporate a gendered analysis of battering into their interventions. This analysis views battering as a tool of male control of women in intimate relationships. This control, and men s belief in their entitlement to control, rather than physical battering alone, becomes the focus of intervention. Attention is focused on male sex role socialization, and how it shapes men s abusive behaviors and the beliefs that underlie those behaviors.

Controversy in the literature has surrounded the use of social learning approaches which are not informed by such a gender analysis ( Gondolf & Russell, 1987; Tolman & Saunders, 1988). However, many programs incorporate both social learning and profeminist content. Studies to date have not isolated these components in terms of effectiveness in ending abusive behavior. While some programs may be explicitly profeminist in approach, without social learning components, no evaluations of such programs have yet appeared in the research literature. The program most widely associated with the profeminist approach, the Domestic Abuse Intervention Project, in Duluth Minnesota, has been evaluated and is discussed below. However, even the Duluth model integrates cognitive-behavioral content into its approach ( Pence & Paymar, 1993).

Structure and Length Although programs vary in length, most are relatively short-term, ranging from 6 to 32 weeks in duration ( Eisikovits & Edleson, 1989; Tolman & Bennett, 1990). Generally, programs are relatively structured in format, though calls for less structured groups have been made ( Jennings, 1987).

Defining Effectiveness

One critical issue in determining effectiveness of intervention with batterers is how success is defined. At one end of the continuum, some researchers have used statistically significant reductions in physical violence to claim program success ( Neidig, 1986). At the other end, others have advocated for nothing short of a transformation of program participants until men are prepared to take social action against the woman-battering culture" ( Gondolf, 1987).

Physical Abuse.Most studies define cessation of physical abuse as the primary criterion for judging effectiveness of efforts. Some studies consider reduction of violent behavior a success, while others set complete cessation of violence as the criterion for success. Viewing reduction as success is questionable because reduction of violence will not necessarily alter the terror that battered women feel as a result of the violence. Even so called minor violence may result in injury or serve as threats of subsequent more injurious violence. Many also agree that ending threats of violence is an important goal. Most program evaluators have attempted to measure the degree to which men s programs end violent behavior, but only a few have examined a range of direct and indirect threats of violence. It is likely that greater attention will be paid to some types of threatening behavior as states and localities adopt laws expanding the definition of illegal actions to include behavior like stalking.

When considering physical abuse and threats of abuse, in addition to what is measured, the source of measurement is critical. Some studies use official indicators, like records of rearrest to determine effectiveness. Police data probably underreport incidence, compared to women s self-reports ( Dutton 1986) because police reports define success as the absence of crime rather than the absence of abuse. Many incidents of physical and psychological abuse might have occurred without anyone calling the police. Batterer self-report is particularly suspect, as male batterers are likely to deny or minimize their abusive behavior and underreport their own abusive acts. Therefore, more confidence can be placed in those studies using victims reports or combined victim-offender-police reports. An index that combines reports of abuse from any source is the most conservative estimate of success in terms of physical abuse ( Hamberger & Hastings, 1988).

The percentage of program participants actually contacted for follow-up presents a critical evaluation issue. If one defines success rate as the percentage of cases known to be nonviolent from the total original sample, then success rates look dramatically different from those reported in many studies. For example, in one evaluation of the Domestic Abuse Project, in Minneapolis, Minnesota ( Edleson & Syers, 1990; 1991) over 500 men initially contacted the agency during a 12 month period. Of these, 283 completed the intake process and attended an initial group session. One hundred fifty three of those who attended the initial group session completed the program. In addition, only approximately two out of three of those men for whom follow-up data were available were reported to be non-violent. These data indicate that thirty percent of the men who initially contacted the agency completed the program to which they were assigned. Fifty-four percent of the men who attended an initial session completed the program. Since two-thirds of those who completed the program were reported non-violent during the follow-ups, in the final analysis, one could argue just one in five (20%) of those who originally contacted the agency made it all the way through the program and were documented to be non-violent during follow-up periods. On the other hand, one can also correctly view the results as demonstrating that of the men who completed treatment and were located, two out of three were non-violent subsequent to treatment.

This example illustrates the need for full information about who is included in study success rates. We do not expect intervention to be effective for those who do not complete it, and certainly not for those who do not choose to participate in it. Consequently, when we attempt to compare studies, and use those data to draw conclusions about the effectiveness of batterers intervention, it is important that we be clear about who is included in the success rates.

Beyond physical abuse and threats of physical harm a number of outcomes might be considered in determining the success of batterers intervention. Figure 1 presents a range of outcomes that might be considered. The rationale for the inclusion of each of these is described briefly below.

Figure 1: Measures of Success in Batterer Intervention

Intervention should reduce or prevent further:

  • Injury to the victim

  • Rearrest or other types of official recividism

  • Physical abuse

  • Psychological maltreatment

  • Sexual abuse

  • Separation abuse

  • Violence supporting attitudes and beliefs held by the batterer

Intervention should increase batterers':

  • Egalitarian partnership

  • Positive behaviors

  • Social Skills

  • Prosocial, antiviolence attitudes

  • Psychological and social functioning

Batterer Intervention should Improve:

  • Survivor's Well-being

    • Reduction of Fear

    • Improved psycho-social functioning and reduction of traumatic stress symptoms

  • Children's Well-being

Sexual Abuse.Finkelhor and Yllo ( 1982) point out that the great majority of rapes are believed to occur between intimate partners or acquaintances. Several studies have found that between 9% and 14% of women report having been forced to perform some type of sexual activity against their will by their male partners ( Hanneke, Shields, & McCall, 1986; Russell, 1982). Batterers may engage in various forms of sexual abuse towards their partners. Sexual abuse may be viewed as a continuum of behaviors ranging from telling sexist jokes or making unwanted or demeaning comments about a partner s body, to demands for sexual activity, to the actual use of physical force or other weapons to rape and force other sexual activity. Some program evaluations have included an evaluation of sexual abuse, though often it has been measured by single items included in indices of physical abuse (cf, Edleson & Syers, 1990; Harrell, 1991; Tolman & Bhosley, 1990). Successful batterer intervention should lead to elimination of sexually abusive behaviors.

Psychological Maltreatment.Beyond physical abuse and threats of violence, batterers may engage in a broad range of abusive and controlling behaviors. Broadly construed, psychological maltreatment can be any behavior that is harmful or intended to be harmful to one s partner. It may be useful to view such maltreatment on a continuum. On the one end are isolated hurtful behaviors that may occur in any relationship: withdrawing momentarily, listening unempathically, speaking sharply in anger. On the other end of the continuum is pervasive, one-sided psychological torture, paralleling intentional brain-washing and mistreatment of prisoners of war ( Andersen, Boulette & Schwartz, 1991). Among batterers, psychological maltreatment frequently, if not always, accompanies physical maltreatment. Theoretically, psychological maltreatment can be viewed as functionally equivalent to physical abuse: both may function to establish dominance and control over another person ( Tolman & Edleson, 1989; Pence, 1989). Therefore, evaluating the extent to which intervention impacts upon these behaviors is important in determining the success of interventions for batterers.

Separation abuse.The risk of violence continues during separation and after divorce, raising concerns about safe custody visitation arrangements (see Saunders, 1994). For example, Minnesota police reported that almost half (47%) of battered women were victimized by an ex-spouse or friend, exceeding the percent of those married to their abuser (44%) ( Minnesota Department of Corrections, 1987). In a Canadian study ( Leighton, 1989) a quarter of the women reported threats against their lives during custody visitations. Abuse characteristic of the separation period may include harassing phone calls or harassment at work, following her around, an using visitation with children to as an excuse to have contact with his partner. These types of behaviors would need to be systematically evaluated, especially as many abusers are separated from their partners during and subsequent to intervention, and may be falsely viewed as successful in changing their behavior if these separation specific behaviors are not evaluated.

Increasing Positive or Caring Behaviors .Given that men in abusers treatment may make important changes in behavior beyond cessation of physical abuse, measurement of positive, caring or egalitarian behavior could be viewed as an important outcome of intervention. Many programs not only focus on how men may stop their abusive behavior, but also seek to increase positive behaviors, e.g. their ability to negotiate fairly, express feelings in an open manner, improve their parenting, and increase their shared responsibility for household responsibilities. The Equality Wheel ( Pence & Paymar, 1993) is one tool used by programs for directing positive change efforts. However, few studies have included measures of such behavioral change in their evaluations. While acquisition of skills is viewed as a positive outcome, it should be noted that evaluation of skill acquisition, without measurement of abusive behavior as an outcome, would not be sufficient for judging success in a program.

Well-being of survivors.Another important category of outcomes relates to the well-being of batterered women. As stated previously, while stopping overt abuse is good, it does not mean that women will not continue to be terrorized or dominated by other forms of abusive behavior. If the ultimate goal of intervention is not just to stop the abuse, but to improve the lives of battered women, then safety, physical and psychological well-being of survivors of abuse must be considered as ultimate outcomes of intervention. Tolman & Bhosley ( 1990), Harrell, ( 1991) and Saunders ( 1994), for example, included measures of women s fearfulness of their partners in their evaluations. Measuring reduction of psychological and physical symptoms of survivors would also enhance our understanding of how interventions aimed at abusers benefit their partners, and which types of intervention lead to the greatest overall well- being of survivors. For example, Tolman & Bhosley ( 1990) examined the psychological well-being of battered women approximately one year after their partners had participated in batterer groups. Psychological maltreatment by batterers toward their partners was found to be significant predictor of the women s well-being, even when physical abuse was controlled for statistically.

Well-being of children.Estimates of the number of children who live in homes where their mothers are also being physically abused vary. For example, child protection workers in the Massachusetts Department of Social Services ( Hangen, 1994) reported statewide that an average of 32.48% of their cases also involved domestic violence. A somewhat higher estimate was obtained in Straus and Gelles 1985 national survey of over 6,000 American families (see Straus & Gelles, 1990). They found that 50% of the men who frequently assaulted their wives also frequently physically abused their children. They also found that mothers who were beaten were at least twice as likely to abuse physically their children than were mothers who were not abused.

A related concern is the degree to which children in a home witness wife assaults and to what extent such witnessing influences social development and mental health. The most widely cited estimates of the number of children who witness violence (but are not themselves abused) comes from the work of Carlson and Straus. Somewhere between 3.3 million ( Carlson, 1984) and 10 million ( Straus, 1991) children in the United States are said to be at risk of witnessing woman abuse each year. Studies have found that children who live with domestic violence in their homes exhibit more aggressive and antisocial (externalized) as well as fearful and inhibited (internalized) behaviors ( Christopherpoulos et al., 1987; Jaffe et al., 1986) and to show lower social competence than other children ( Wolfe, 1986). Children who witnessed abuse were also found to show more anxiety, depression and temperament problems ( Christopherpoulos et al., 1987; Forsstrom-Cohn &Rosenbaum, 1985; Holden & Ritchie, 1991; Hughes, 1988; Westra & Martin, 1981), than children who did not witness violence at home. There is also some support for the hypothesis that children from violent families of origin carry violent and violence-tolerant roles to their adult intimate relationships ( Cappell & Heiner, 1990; Rosenbaum & O Leary, 1981; Widom,1989).

To date, the well-being of children subsequent to batterer intervention has not been studied. If batterer intervention is successful, then improvements for their children might also be a predictable outcome. The improvement of parenting skills of batterers, increasing maternal well-being, as well as cessation of abuse that children witness could all contribute to better functioning for children subsequent to batterer intervention.

Measuring couples functioning.In our view, the primary goal of batterer intervention is not to improve relationships, but rather to end the abusive behavior. However, ending abusive behavior and increasing use of positive and egalitarian relationship behaviors of course should lead to more positive functioning of relationships. Improved couple functioning for those couples who choose to stay together subsequent to the cessation of abuse would be more likely to occur for those who not only end abuse, but improve in these other relationship skills. On the other hand, successful intervention for batterers might in some cases, increase the probability of separation. Some battered women report they feel safer to leave when their partners have begun to demonstrate their choice to refrain from abusive behavior. Therefore, it is important not to judge the success of batterer intervention by measuring improvement in couple functioning, as a desirable goal is freeing women from coercive constraints to remain in their relationships.

An ultimate goal of batterer intervention for those who separate might be termed empowered autonomy" of the survivor. Ideally, survivors should feel safe from subsequent abuse, and unhindered by actions of their abuser. If batterers have taken responsibility for their actions, and responded accountably and empathically to their partners, survivors will be batter able to heal from the abuse and continue their lives.

Other mediators-attitude change, mental health, symptoms.Batterer intervention may seek to produce a number of changes in men that are believed to be important in ending abusive behavior. One goals is change attitudes and beliefs that are believed to lead to abusive behavior (e.g. denial of abuse or responsibility for it, entitlement to control of women, male superiority) and improve support for attitudes and beliefs that enhance non-violent or supportive behavior (e.g. belief in gender equality, endorsement of non-violent conflict resolution strategies). In addition, reduction of mental health symptoms or personality change ( Hamberger & Hastings, 1989), reduction or cessation of use of alcohol or other drugs, and numerous other behaviors, symptoms and cognitions (cf Saunders & Hanusa, 1986; Maiuro et al, 1987) might be targets of change in various forms of batterer intervention. Measurement of these variables can lead to improved knowledge mediators of change in abusive men. While studying changes in these presumed mediating variables are important, it must be emphasized that program success ultimately should be judged by criteria directly related to the abuse, rather than these mediating variables. Reconciling Evaluation Goals. We have reviewed a broad range of outcomes that could be considered in evaluation of the success of batterers intervention. Having raised these potential goals, it is important to return to the question of what can reasonably be considered as the criteria for judging success. While consensus is clear on ending physical abuse and threats of abuse, it is not clear what we can expect beyond those goals. Can we expect such programs to end men s use of manipulative or sexist behavior that is characteristic of many intimate relationships where violence does not occur? Some critics of batterers programs would likely argue that these programs must be an active part of the transformation of male-female relationships and at present, they fall far short of such a vision. On the other hand, many policy makers and most funders of batterers programs are more concerned with ending illegal, violent behavior regardless of the degree to which the participants are transformed. In between these two positions are those who recognize that power and control may well be maintained even after violence and threats of violence have ended. However, they may question the appropriateness of expecting or mandating through the courts that certain men identified by social agencies attend a program to become transformed when others, who are not violent, but who may be regularly applying unfair power and control in their own relationships, are not also required to attend programs and make similar changes. Given these difficult issues, we should make it clear that in our own work, we have advocated that batterer intervention include a focus on behavior change techniques for stopping physical and other forms of abuse, as well as on promoting social change efforts by program participants ( Edleson & Tolman, 1992).

In summary, the ways in which one defines success in batterers programs will greatly influence one s conclusions about their effectiveness. There is broad support only for only for a limited social mandate for batterer intervention to assist men in ending the use of illegal violent and threatening behaviors. Unfortunately, there is not yet a consensus of support for large-scale social interventions aimed at changing many non-violent men s controlling behaviors toward women. If that consensus is ever reached, it is unlikely that psychoeducational groups aimed at a targeted subgroup of men will be the primary means for bringing about the desired social change in all men s behavior toward women. Broader preventive and social change efforts will undoubtedly be necessary.

Research on Groups for Men Who Batter

A number of other authors have extensively reviewed research on the effectiveness of batterer intervention ( Saunders & Azar; 1989; Hamberger & Hastings, 1993; Rosenfeld, 1992). Since the publication of these reviews, only a few new studies have appeared. Our own reviews ( Tolman and Bennett, 1990; Eisokovits & Edleson, 1989) revealed consistent findings that in varying programs, using various methods of intervention, a large proportion of men stopped their physically abusive behavior subsequent to involvement in the programs. When reports of physical abuse based on partner reports are considered, successful outcomes ranged from 53% to 85%. In these studies, follow-up ranged from 4 months to 26 months. On the whole, evidence suggests that many batterers are successful in stopping direct physical abuse for at least a short-time subsequent to intervention. However, a longitudinal study of men in the general population also found high rates of cessation of physical abuse, even though there had been no formal intervention ( Feld & Straus, 1990). Consequently, it is unclear whether treatment itself is the active ingredient in accounting for the differences observed between treated and untreated offenders. The favorable evidence for effectiveness of batterers treatment must be weighed in light of other explanatory factors and with a high degree of caution due to many methodological shortcomings evident in the studies thus far reported. For example, lower percentages of success tended to occur in programs with lengthier follow-up and when success was based on report of women victims rather than official arrest rates or men s self-reports. In addition, many social service programs experience high levels of offender attrition, further diluting their overall effectiveness (see Deschner, 1984; Harrell, 1991).

Table 1 summarizes studies using partner reports as outcomes . A major methodological shortcoming of the existing literature on group treatment is the scarcity of experimental studies, leaving open the question of whether intervention itself is responsible for change in abusive men s behavior. Some quasi-experimental studies have been completed, generally using men who drop out of treatment as comparisons for those who do complete. Several of these studies have indicated that batterers groups are effective for men who complete treatment, when compared to quasi-experimental controls who begin but do not complete treatment (see Edleson & Grusznski, 1988). However, even in these studies, there are relatively small differences in success rates between completers and non-completers.

Table 1. Table 1: Physical Abuse Recidividsm Rates for Batterers Intervention Groups Based on Partner Reports

First Author/YearRecividism Rate for Trated Group (theoretical base and or format)Recidivism Rate for Comparison GroupFollow-up Period
Dutton, 198616% (cognitive-behavioral)40% (untreated)6 months to 3 years
Edleson, 1988-133% (Cognitive-behavioral, profeminist)46% (non-completers)4.6 months ave
Edleson, 1988-232% (cognitive-behavioral, profeminist) 4.6 months ave
Edleson, 1988-341% (cognitivie-behavioral, profeminist)48% (non-completers)4.6 months ave
Edleson, 199045.9% (32 session) 34.5% (12 session) 32% (education) 54% (self-help) 34% (combined) 6 months
Edleson, 199140% (12 session) 23.3% (32 session) 36.4% (education) 37.0% (combined) 21.1 % (self-help 18 months
Hamberger, 198828% (cognitive-behavioral)47% (non-completers(12 months
Hamberger, 199030% (cognitive-behavioral) 12 months
Harrell, 199143% (diverse, three different programs)12% (untreated)4 months
Poynter, 199135% (cognitive-behavioral; profeminist) 12 months
Saunders, 199440% (cognitive-behavioral) 40% (psychodynamic) 12-18 months
Shepard, 198731% (profeminist, educational) 14 months
Shepard, 199240% (profeminist, educational) 5 years
Tolman, 198747% (cognitive-behavioral, profeminist ave, 26 months
Tolman, 199142% (cognitive-behavioral, profeminist) 12 months

Several recent studies warrant specific consideration . Harrell ( 1991) found that batterers who completed short-term court-mandated groups were as likely to commit subsequent physical abuse as men who were found guilty by the court but were not mandated to treatment. This study was important for several reasons. One strength of the study was its use of a quasi-experimental control group of untreated men. As stated above, past studies have tended to use only quasi-experimental comparisons of treatment completers and drop-outs. In addition, Harrell attempted to control for differences in treated and untreated men by adjusting for variables found to be different for the treated and untreated groups. These included employment, prior criminal history, and living with the victim. A further strength was the richness of outcome data gathered. Not only were physical abuse and threats of violence measured, but also psychological abuse, conflict resolution skills, beliefs about wife abuse, and the victims perceived safety. Reports on recidivism were derived from partner reports, rather than self-report. Official recidivism data was also collected from police and court records. In addition, a number of modifiable risk factors and mediators of violence were measured to see if intervention resulted in desired changes on these factors. The results on these various measures were consistent: those who received the batterer groups did not do any better than those who did not.

Harrell s findings that treated men did worse on the measure of physical aggression than untreated men are surprising and sobering. However, it would be incorrect to conclude on the basis of this one study that treatment is ineffective or harmful, though these possibilities must be considered. Harrell notes several possible explanations for the results. Among the most important considerations was that the treatment models or implementation of the models was inadequate. The programs evaluated were relatively short-term (12 sessions) and not intensive (1 to 1.5 hours per session). The follow-up period, 4 months across the treatment period, may not have been long enough to capture the longer term impact of the programs. However, this would not explain why the treated group was more likely to recidivate rapidly, when the untreated group was not. One would be more likely to expect the opposite.

Another important issue that Harrell raises is that despite the statistical controls and the belief that the court did not exercise any systematic bias in assigning men to treatment, it is possible that unmeasured differences between the treated and untreated controls bias the analysis towards the untreated group. When comparing the recidivism rates obtained in the Harrell study with the others reported in Table 1, what becomes clear is that the rates of recidivism for the untreated group were remarkably low. The treated groups recidivism were not exceedingly high, although they are relatively high for a short follow-up period. What is most difficult to understand about these findings is not the poor performance of the treated men, but the low recidivism of the men not referred to treatment.

Palmer et al. ( 1992) conducted the only study to date which used random assignment of batterers to treatment and a no-treatment control group. Thirty court-ordered batterers were assigned to a 10 session group which met for 1.5 hours per week. The program could be characterized as psychoeducational, with cognitive-behavioral content as well as a focus on sex roles. Although the researchers attempted to get partner reports of abusive behavior, they were unsuccessful in gaining enough for group comparisons. Based on police reports, only 10% of the treated batterers recidivated, while 30% of the untreated controls were recidivists.

In light of the Palmer results, it is interesting to note that in the Harrell study, using new arrests, rather than calls to the police as the criterion, 19% of the treated group compared to 7% of the untreated group would be considered recidivists. This highlights that Harell s findings are contradictory to those of Palmer, as those of Syer & Edleson ( 1991) and Steinman ( 1990), who found mandates to treatment were associated with lower, not higher recidivism using official data as the criterion. These mixed results indicated that we must look further at intervention in the court context before drawing conclusions about efficacy of intervention.

Edleson and Syers ( 1990; 1991) randomly assigned batterers to different treatment formats. Men were randomly assigned to two differing intensities of treatment. Some men were assigned to a program of 12 sessions over 3 months while others were assigned to receive a 32 session program over 4 months. Within each of the two differing intensities, men were assigned to receive either (1) a highly structured, educational program; 2) an unstructured self-help format; or 3) a program that combined self-help group processes with education. They found that approximately 2 out of 3 men assigned to receive structured educational programs or those combining education with group processing were reported not violent by their partners during 6 and 18 month follow-up periods. Men in these groups achieved more stable outcomes than did those randomly assigned to a less structured, self-help program. The study also found that 12 session programs achieved outcomes similar to 32 session program. Factors including a participant s education and involvement with the court also influenced outcomes. The higher a man s education the more likely he would be reported not violent at follow-up. A man s involvement with the court system statistically predicted his non-violent outcome at the 18-month follow-up but not at the 6 month follow-up.

Saunders ( 1994) conducted the first study to utilize random assignment to theoretically distinct comparison groups. He compared the efficacy of cognitive- behavioral, profeminist group treatment with a process-psychodynamic group. Each group met for 2 1/2 weekly sessions, over a 20 week period. There was no advantage for either treatment, based on partner and official reports. Saunders analysis revealed however that there were significant interaction effects. Most notably, men with dependent personalities, as measured by the Millon Clinical Multiaxial Inventory (MCMI), were more successful in the process psychodynamic condition.

Studies On Couples Groups And Couples Treatment

While most studies to date have focused on groups for batterers, there has been some empirical evaluation of conjoint couples counseling and multi-couple groups interventions. Previous studies have been reviewed in detail elsewhere ( Edleson & Tolman, 1992; Saunders & Azar, 1989) and no newer studies have been recently published. Therefore, only a brief review of this literature follows.

Research on couples intervention, like batterers groups, has largely focused on programs with cognitive-behavioral elements. Several authors have reported evaluation data for conjoint counseling done with individual couples ( Lindquist et al, 1983; Taylor, 1984) and multi-couple groups ( Neidig et al, 1985; Deschner & McNeil, 1986).

The Lindquist study reported a recidivism rate for physical abuse of 100%. All of the eight couples reported in the study reported subsequent violence after a six-month period. Taylor ( 1984) briefly reported that of 50 couples he worked with in treatment, 35% reported new violence in the following six-months However, Taylor does not state how these data were collected, nor who reported them.

Using multi-couple groups, the Neidig and Deschner studies reported more success for cessation of physical violence, with recidivism rates of 13% and 15%, respectively. However, drawing conclusions about the effectiveness of couples approaches from these studies is hampered for several reasons.

Neidig et al ( 1985) evaluated a highly structured 10 session skill building approach used with groups of couples. Their report did not state who reported the recidivism rate of 13% at four-month follow-up was determined, nor who was covered by the report. If the figure includes non-violent partners, then the recidivism rate underestimates the actual level of violence.

Deschner ( 1984) reports that in an eight-month follow-up of 15 couples who completed a 10 session cognitive-behavioral group, 8 were violence free. Five couples who participated in their program later experienced up to four minor physical incidents such as grabbing or slapping the partner" (p. 60). Deschner considers all of these thirteen couples (87%) -including the five in which there were minor incidents- to be free of battering. Given our previous discussion, this definition of success is questionable, and highlights the need for clear definitions of successful outcome when interpreting data from studies.

Future Directions and Conclusion

Recent studies on batterer intervention advance our understanding of the potential benefits and limitations of our current responses. We find that the research on criminal justice system response does not yield any conclusive answers about how we might best respond to male violence. Research, however well-executed, rarely provides clear-cut direction in policy and practice. Rather, research studies become part of an ongoing dialogue about how to best meet the challenge of social problems. In this light, we see that arrest alone is not always a better deterrent than non-response, nor is it any worse than other alternatives. On the contrary, when paired with other system responses like prosecution and mandates to counseling, there is evidence that coordinated responses by communities can be a deterrent to subsequent violence by men towards their female partners. Even there some contradictory evidence confronts us at present. In contrast to several other studies, Harrell's ( 1991) study of batterers who were mandated and not mandated to intervention found no positive effect for treatment, while other studies have demonstrated positive effects for intervention with similar comparisons ( Palmer, 1992; Dutton, 1986). Further research is not likely to yield unequivocal answers, but will undoubtedly help us refine our understanding of how these interventions contribute to change.

The most recent studies have substantially improved the rigor of quantitative methods used to evaluate interventions. Studies are beginning to move away from pre-post or post-hoc evaluations of single programs and towards randomized comparisons of different interventions, or intervention structures. Studies such as Saunders ( 1994) hold particular promise for moving away from the one size fits all" stage for intervention. Most programs for men who batter run standardized curricula in which a diverse group of men participate. Programs seldom offer tracking of particular subgroups of men e.g. first-time offenders, men reported violent with only their intimate partners, men who are generally violent, or men with significant mental health problems. Research that has attempted to identify typologies of men who batter ( Dutton; 1988; Gondolf, 1988; Saunders, 1992, 1993) might lead to more specific programming based on these classifications. There also needs to be an improvement in culturally competent programming for men of color ( Williams, 1992). Agency structures and program curricula must be refined so that they effectively address men of color, who are more likely than other men to drop out of current programs ( Saunders & Parker, 1989). The more we learn about the factors which impact upon intervention success, and how they differ among batterers, the better we will become at matching interventions to specific men, and improving our success at changing their behavior.

Even the limitations demonstrated by current research are valuable. The modest and somewhat contradictory patterns of outcomes suggests that we should not direct all our efforts towards criminal justice or psychoeducational intervention for batterers. We must consider the need for broader intervention and prevention efforts. We look forward to future research that will add useful data to the dialogue about how to best end male violence against women.


Adams & McCormick, A. (1982). Men unlearning violence: A group approach. In M. Roy (Ed.), The abusive partner: An analysis of domestic battering(pp. 170-197). New York: Van Nostrand Reinhold.

Anderson, S. M., Boulette, T.R. & Schwartz, A. H. (1991). Psychological maltreatment of spouses. In R.T. Ammerman & M. Hersen (Eds.), Case studies in family violence.New York: Plenum Press.

Buzawa, E.S. & Buzawa, C. (1990). Domestic violence: The criminal justice response.Newbury Park, CA: Sage.

Cahn, N. R. (1992). Innovative approaches to the prosecution of domestic violence crimes: An overview. (Pp.161-178) in E.S. Buzawa & C.G. Buzawa (Eds.) Domestic violence: The changing criminal justice response,Westport, CT: Greenwood Press.

Cook, D. R. & Frantz-Cook, A. (1984). A systemic approach to wife battering. Journal of Marital and Family Therapy,10, 83-93.

Coleman, K.H. (1980). Conjugal violence: What 33 men report. Journal of Marital and Family Therapy,6, 207-213.

Deschner, J. (1984). The hitting habit: Anger control for battering couples.New York: Free Press.

Deschner, J.P. & McNeil, J.S. (1986). Results of anger control training for battering couples. Journal of Family Violence,1, 111-120.

Dunford, F. W., Huizinga, D. & Eliot, D.S. (1990). The role of arrest in domestic assault: The Omaha police experiment, Criminology28: 183- 206.

Dutton, D. G. (1987). The criminal justice response to wife assault. Law and Human Behavior11: 189-206.

Dutton, D. G. (1988). The domestic assault of women: Psychological and criminal justice perspectives.Boston: Allyn and Bacon.

Dutton, D. G. (1986). The outcome of court-mandated treatment for wife assault: A quasi-experimental evaluation. Violence and Victims1: 163-175.

Edleson, J. L. & Syers, M. (1990). The relative effectiveness of group treatments for men who batter. Social Work Research and Abstracts26: 10-17.

Edleson, J. L. & Syers, M. (1991). The effects of group treatment for men who batter: An 18 month follow-up study. Research in Social Work Practice,1, 227-243.

Edleson, J. L. & Grusznski, R. J. (1988). Treating men who batter: Four years of outcome data from the domestic abuse project. Journal of Social Science Research,12, 3-22.

Edleson, J.L. & Tolman, R. M. (1992). Intervention for men who batter: An ecological approach.Newbury Park: Sage Publications.

Eisokovits, Z. C. & Edleson, J. L. (1989). Intervening with men who batter: A critical review of the literature. Social Service Review,pp. 384-414.

Feazell, C. S., Mayers, R.S. & Deschner, J.P. (1984). Services for men who batter: Implications for programs and policies. Family Relations,33, 217- 223.

Ferraro, K. & Boychuk, T (1992). The court s response to interpersonal violence: a comparison of intimate and non-intimate assault.In E.S. Buzawa and C. Buzawa (Eds.) Domestic violence: The criminal justice response,(pp. 209-225). Westport,CT: Greenwood.

Ferraro, K. & Pope, L. (1993). Irreconcilable differences: Battered women, police and the law. In N.Z. Hilton (ed.) in Legal responses to wife assault: Current trends and evaluation,(Pp.96-126). Newbury Park, CA: Sage.

Ford, D. (1983). Wife battery and criminal justice: a study of victim decision- making. Family Relations32:463-475.

Ford, D. & Regoli, M.J. (1992). The preventive impact of policies for prosecuting wife batterers. In E.S. Buzawa and C. Buzawa (Eds.) Domestic violence: The criminal justice response,(pp. 181-207). Westport,CT: Greenwood.

Geller, J.A. (1982). Conjoint therapy: Staff training and treatment of the abuser and abused. In M. Roy (Ed.) The abusive partner(pp. 155-173). New York: Springer.

Geller, J. A. & Walsh, J. (1997-1978). A treatment model for the abused spouse. Victimology,1, 627-632.

Gondolf, E.W. (1987). Changing men who batter: A developmentalmodel for integrated interventions. Journal of Family Violence,2, 335-350.

Hamberger, L. K. & Hastings, J.E. (1988). Skills training for treatment of spouse abusers: An outcome study. Journal of Interpersonal Violence,1, 363- 373.

Hamberger, L. K. & Hastings, J.E. (1993). Court-mandated treatment of men who assault their partner: Issues, controversies, and outcomes In Z. Hilton (Ed.) Legal responses to wife assault(pp. 188- 229). Newbury Park: Sage.

Hamberger, L. K., & Hastings, J. E. (1988, August). Recidivisim following spouse abuse abatement counseling: Treatment program implications.Paper presented at the meeting of the American Psychological Association, Atlanta.

Hamberger, L. K. & Hastings, J.E. (1990). Different rotuse to mandated spouse abuser counseling II: A cross-validation study.

Harrell, A.V. (1991). Evaluation of court-ordered treatment for domestic violence offenders.Final report submitted to the State Justice Institute. Washington, DC: The Urban Institute.,

Hart, B. (1993). Battered women and the criminal justice system. American Behavioral Scientist36:624-638.

Hirschel, J. D., Hutchison, 3rd, , I. W. & Dean, C.W. (1992). The failure of arrest to deter spouse abuse. Journal of Research In Crime And Delinquency ,29: 7-33.

Jaffe, P., Hastings, E. Reitzel, D. Austin, G. (1993). The impact of police laying charges. In N.Z. Hilton (ed.) in Legal responses to wife assault: Current trends and evaluation,(Pp 62-95). Newbury Park, CA: Sage.

Jennings, J. L. (1987). History and issues in the treatment of battering men: A case for unstructured group therapy. Journal of Family Violence,2, 193-214.

Lindquist, C. V., Telch, C. F. & Taylor, J. (1984). Evaluation of a conjugal violence treatment program: A pilot study. Behavioral Counseling and Community Intervention,3, 76-90.

Mauiro, R. D., Cahn, T.S., Vitaliano, P.P. & Zegree, J.B.(1987, August). Treatment for domestically violent men: Outcome and follow-up data.Paper presented at the meeting of the American Psychological Association, New York.

Margolin, G. (1979). Conjoint marital therapy to enhance anger management and reduce spouse abuse. American Journal of Family Therapy,7, 13- 24.

Millon, T. (1983). Millon Clinical Multiaxial Inventory.Minneapolis: National Computer Systems.

Neidig, P. H., & Friedman, D. H. (1984). Spouse Abuse: A Treatment Program for Couples.Champaign, IL. Research Press.

Neidig, P.H. , Friedman, D. H. & Collins, B.S. (1985). Domestic Conflict Containtment: A spouse abuse treatment program. Social Casework,66, 195-204.

Neidig, P.H. (1986). The development and evaluation of a spouseabuse treatment program in a military setting. Evaluationand Program Planning,9, 275-280.

Palmer, S. E., Brown, R.A., Barrera, M.E. (1992). Group treatment program for abusive husbands: Long term evaluation. American Journal of Orthopsychiatry62(2), 276-283.

Pence, E. (1983). The Duluth Domestic Abuse Intervention Project. Hamline Law Review,6, 247-275.

Pence, E. (1990). Batterer programs: Shifting from community collusion to community confrontation. In P.L. Caesar & L.K.Hamberger (Eds.), Treating men who batter.NY: Springer.

Pence, E. & Paymar, M. (1993). Education groups for men who batter.New York: Springer.

Pirog-Good, M. & Stets-Kealey, J. (1985). Male batterers andbattering prevention programs: A national survey. Response,8, 8-12.

Poynter, T.L. (1991). An evaluation of a group programme for male perpetrators of domestic violence: A follow-up study. Australian Journal of Marriage & Family,12 (2), 64-76.

Rosenfeld, B. D. (1992). Court-ordered treatment of spouse abuse. Clinical Psychology Review,12, 205-226.

Saunders, D. G., & Azar, S. (1989). Family violence treatment programs: Descriptions and evaluation. In L. Ohlin & M. Tonry (Eds.), Family violence: Crime and justice, a review of research(pp. 481-546). Chicago, IL: University of Chicago Press.

Saunders, D. G., & Hanusa, D. R. (1986). Cognitive-behavior treatment of men who batter: The short-term effect of group therapy. Journal of Family Violence,1(4), 357-372.

Saunders, D. G., & Parker, J. C. (1989). Legal sanctions and treatment follow-through among men who batter: A multivariate analysis. Social Work Research and Abstracts,(September, 1989), 21-29.

Saunders, D. G. (1994, November). Cognitive-behavioral and process- psychodynamic treatments for men who batter: Interactions between offender traits and treatments.Paper presented at the Association for the Advancement of Behavior Therapy.

Schecter, S. (1982). Women and male violence: The visions and struggles of the battered women's movement.Boston: South End.

Shepard, M. (1992). Predicting batterer recidivism five years after community intervention. Journal of Family Violence,7 (3), 167-178.

Shepard, M. (1987, July ). Intervention with men who batter: An evaluation of a domestic abuse program.Paper presented at the Third National Family Violence Conference, University of New Hampshire, Durham.

Sherman, L. W. (1992). The influence of criminology on criminal law: Evaluating arrests for misdemeanor domestic violence. The Journal of Criminal Law & Criminology,83: 1-45.

Sherman, L. W. (1992). Policing domestic violence: Experiments and dilemmas.New York: Free Press.

Sherman, L. W. & Berk, R.A. (1984). The specific deterrent effects of arrest for domestic assault. American Sociological Review49: 261-272.

Sherman, L.W., Schmidt, J.D. ,Rogan, J.P. Gartin, P.R. Chon, E.G. Collins, D.J. & Bacich, A.R. (1991). From initial deterrence to long-term escalation: Short-term custody for poverty ghetto domestic violence. Criminology29: 821-849.

Soler, E. (1987). Domestic violence is a crime: A case study-San Francisco Family Violence Project. In D. J. Sonkin (Ed.), Domestic violence on trial: Psychological and legal dimensions of family violence(pp. 21-28). New York: Springer.

Steinman, M. (1988). Evaluating a system-wide response to domestic violence: Some initial findings. Journal of Contemporary Criminal Justice4: 172-186.

Steinman, M. (1990). Lowering recidivism among men who batter women. Journal of Police Science and Administration17: 124-132.

Syers, M. & Edleson, J.E. (1992). The combined effects of coordinated criminal justice intervention in woman abuse. Journal of Interpersonal Violence7: 490- 502.

Taylor, J. W. (1984). Structured conjoint therapy for spouse abuse cases. Social Casework,65, 11-18.

Tolman, R. M., Beeman, S. & Mendoza, C. (1987, July). The effectiveness of a shelter-sponsored program for men who batter: Preliminary results.Paper presented at the Third National Conference for Family Violence Researchers, University of New Hampshire, Durham.

Tolman, R. M. & Bennett, L. W. (1990). A review of research on men who batter. Journal of Interpersonal Violence,5, 87-118.

Tolman, R.M., & Bhosley, G. (1991). The outcome of participation in a shelter- sponsored program for men who batter. In D. Knudsen & J. Miller (Eds.), Abused and Battered:Social and Legal Responses.Aldine de Gruyter, New York.